Multisample Flow Matching: Straightening Flows with Minibatch Couplings
This work addresses a bottleneck in generative modeling for researchers and practitioners by offering a more efficient method with incremental improvements over existing flow matching techniques.
The paper tackles the problem of training continuous-time generative models by proposing Multisample Flow Matching, a framework that uses couplings between data and noise samples to reduce gradient variance and obtain straighter flows, resulting in improved sample consistency on downsampled ImageNet and better low-cost generation.
Simulation-free methods for training continuous-time generative models construct probability paths that go between noise distributions and individual data samples. Recent works, such as Flow Matching, derived paths that are optimal for each data sample. However, these algorithms rely on independent data and noise samples, and do not exploit underlying structure in the data distribution for constructing probability paths. We propose Multisample Flow Matching, a more general framework that uses non-trivial couplings between data and noise samples while satisfying the correct marginal constraints. At very small overhead costs, this generalization allows us to (i) reduce gradient variance during training, (ii) obtain straighter flows for the learned vector field, which allows us to generate high-quality samples using fewer function evaluations, and (iii) obtain transport maps with lower cost in high dimensions, which has applications beyond generative modeling. Importantly, we do so in a completely simulation-free manner with a simple minimization objective. We show that our proposed methods improve sample consistency on downsampled ImageNet data sets, and lead to better low-cost sample generation.